Time-Frequency Coherence and Forecast Analysis of Selected Stock Returns in Ghana Using Haar Wavelet
Rhydal Esi Eghan *
Department of Mathematics, Kwame Nkrumah University of Science and Technology, Ghana.
Peter Amoako-Yirenkyi
Department of Mathematics, Kwame Nkrumah University of Science and Technology, Ghana.
Akoto Yaw Omari-Sasu
Department of Mathematics, Kwame Nkrumah University of Science and Technology, Ghana.
Nana Kena Frimpong
Department of Mathematics, Kwame Nkrumah University of Science and Technology, Ghana.
*Author to whom correspondence should be addressed.
Abstract
Aims/ objectives: The study seeks to analyze the correlation of some selected stock returns with respect to both time and frequency domain, and also to forecast returns using Wavelet Coherence and Wavelet-ARIMA model as alternative to Pearson correlation and ARIMA model respectively.
Study Design: Financial Mathematics.
Place and Duration of Study: August 2016 to July 2017 , Department of Mathematics, Kwame Nkrumah University of Science and Technology.
Methodology: We transform data using the Haar Wavelet as the basis function.
Results: Results revealed interesting dynamics of correlations altering in time and across frequencies continually between paired returns. Furthermore, Wavelet-Arima method was found to be more appropriate for forecast with minimal error measure of forecast values.
Conclusion: Given the heterogeneous trading behavior in stock markets, investors operate at different frequencies for their trade and investment preferences. Thus, apart from the time domain, there is a frequency domain, which represents various investment horizons.
Keywords: Co-movement, stock returns, wavelet coherence.